<i>Currently, most IoT data are not used. For example, on an oil rig that has 30,000 sensors, only 1 percent of the data are examined. That’s because this information is used mostly to detect and control anomalies—not for optimization and prediction, which provide the greatest value.</i><p>There must be a wealth of data in the other 99%, but you'd have to filter out a bit of noise, preferably closer to the sensor than to some central server so you don't clog the internet with unimportant traffic.